Meta AI's 500M Users Drive Ad Revenue Surge Amid $40B AI Investment and Energy Challenges
October 31, 2024However, Meta is grappling with engineering challenges and high energy demands in the development of its Llama 4 AI model, with executives remaining tight-lipped about energy access issues in the U.S.
The energy requirements are substantial, as a cluster of 100,000 H100 chips demands 150 megawatts of power, which is significantly higher than the 30 megawatts needed for the El Capitan supercomputer, the largest in the U.S.
Meta AI, leveraging its Llama models, boasts over 500 million monthly users and is looking to monetize this growth through advertising as user queries diversify.
Despite facing a 9 percent rise in operating costs, Meta's sales, largely driven by advertising, have surged over 22 percent, leading to increased profits even as the company invests heavily in AI development.
To support its ambitious AI initiatives, Meta plans to invest up to $40 billion in capital expenditures in 2024, representing a significant 42 percent increase from the previous year.
While Meta promotes an open-source strategy for AI, this approach has raised concerns among experts regarding the potential misuse of powerful AI models, although the Llama models are designed to mitigate such risks.
Mark Zuckerberg advocates for the open-source model, asserting that it provides developers with a cost-effective, customizable, and reliable alternative to proprietary systems.
In the broader AI landscape, OpenAI is reportedly training GPT-5, a successor to ChatGPT, which CEO Sam Altman claims will represent a significant leap forward, although specifics about the training cluster remain under wraps.
Additionally, Google is also in the process of developing a new version of its Gemini generative AI models, as confirmed by CEO Sundar Pichai.
Summary based on 1 source